[+cc pkg-systemd-maintainers] On Tue, Apr 14, 2015 at 8:59 PM, Ian Campbell <i...@debian.org> wrote:
> On Tue, 2015-04-14 at 10:02 +0200, Michael Stapelberg wrote: > > > > Apr 02 12:08:28 hostname systemd[1]: Started LSB: Start > > qcontrol daemon. > > > > > > This line in the logfile indicates that systemd is using the sysvinit > > script (hence the “LSB: ” prefix) of qcontrold, not the native systemd > > service file. > > Ah, that seems obvious now you point it out. > > > The git log shows that 0.5.2 is the first version that is supposed to > > have systemd service files, but actually, when I look > > at https://packages.debian.org/jessie/armhf/qcontrol/filelist, I don’t > > see any files in /lib/systemd/system/, which suggests that the Debian > > package doesn’t install the systemd service files that upstream > > provides. > > > > > > Ian, it seems like you should be following > > > https://wiki.debian.org/systemd/Packaging#Using_debhelper_with_dh_systemd > to get the package to properly pick up the systemd service files, at which > point this bug should be fixed. > > I could have sworn this got done (I even thought the patches had come > from you!) but it really hasn't. > > Unfortunately I think switching over to the systemd initfiles for Jessie > is out of the question at this stage, given > https://lists.debian.org/debian-devel-announce/2015/03/msg00016.html. > > Once Jessie is out I'll get the proper fix into Stretch and > Jessie-backports as soon practical. > > In the meantime is there any simple workaround you can think of which > would make the existing sysvinit script work properly with systemd's LSB > compat mode? > > FYI qcontrold.init currently has: > # Required-Start: $local_fs $remote_fs $syslog > > I'd have thought that $remote_fs would be sufficiently late to ensure > the /etc/modules.conf had been processed and udev to have been settled. > But perhaps some other dependency is required? Any idea what it might > be? > pkg-systemd-maintainers, see above: is there a way to wait for udev devices to be available? What’s the current take on udev-settle, would that be a reasonable workaround for this situation? > > systemd-sysv-generator seems like it might be related, but having looked > at the code I don't think it supports anything which would be useful > here. > > Ian. > > -- Best regards, Michael